This commentary evaluates the evidence presented by Alpert, Brown, and Courtois in the Final Report of the APA Working Group on Investigation of Memories of Childhood Abuse. The weaknesses in their assessment of evidence and their theoretical arguments go beyond those identified by Ornstein, Ceci, and Loftus. Historical and clinical data are uncritically overvalued, and empirical concepts (like state-dependent memory) are borrowed for explanatory purposes without appreciating the limits and inconsistencies in the underlying empirical research. An argument is advanced that clinical evidence can never be free of the influence of cherished beliefs and that some therapeutically recovered memories may reflect the cherished beliefs of clinicians who recover them.